- test and fix receiving text/html attachment in multipart/mixed situations

They are now preserved as attachment, instead of diving into parsing-html
  and simplifying.

- adapt mime-debugging
This commit is contained in:
holger krekel
2019-12-07 21:07:53 +01:00
parent b7a9f73329
commit 119d4fa689
5 changed files with 193 additions and 93 deletions

View File

@@ -47,12 +47,10 @@ pub fn dc_receive_imf(
server_uid,
);
/*
if std::env::var(crate::DCC_MIME_DEBUG).is_ok() {
if std::env::var(crate::DCC_MIME_DEBUG).unwrap_or_default() == "2" {
info!(context, "dc_receive_imf: incoming message mime-body:");
println!("{}", String::from_utf8_lossy(imf_raw));
}
*/
let mime_parser = MimeParser::from_bytes(context, imf_raw);
let mut mime_parser = if let Err(err) = mime_parser {

View File

@@ -247,13 +247,13 @@ fn decrypt_if_autocrypt_message<'a>(
// Errors are returned for failures related to decryption of AC-messages.
let encrypted_data_part = match wrapmime::get_autocrypt_mime(mail) {
Err(err) => {
// not a proper autocrypt message, abort and ignore
info!(context, "Not an autocrypt message: {:?}", err);
Err(_) => {
// not an autocrypt mime message, abort and ignore
return Ok(None);
}
Ok(res) => res,
};
info!(context, "Detected Autocrypt-mime message");
decrypt_part(
context,

View File

@@ -2,7 +2,7 @@ use std::collections::{HashMap, HashSet};
use deltachat_derive::{FromSql, ToSql};
use lettre_email::mime::{self, Mime};
use mailparse::MailHeaderMap;
use mailparse::{DispositionType, MailHeaderMap};
use crate::aheader::Aheader;
use crate::blob::BlobObject;
@@ -22,6 +22,7 @@ use crate::param::*;
use crate::peerstate::Peerstate;
use crate::securejoin::handle_degrade_event;
use crate::stock::StockMessage;
use crate::{bail, ensure};
#[derive(Debug)]
pub struct MimeParser<'a> {
@@ -110,7 +111,7 @@ impl<'a> MimeParser<'a> {
mail_raw = raw;
let decrypted_mail = mailparse::parse_mail(&mail_raw)?;
if std::env::var(crate::DCC_MIME_DEBUG).is_ok() {
info!(context, "dc_receive_imf: incoming message mime-body:");
info!(context, "decrypted message mime-body:");
println!("{}", String::from_utf8_lossy(&mail_raw));
}
@@ -509,100 +510,63 @@ impl<'a> MimeParser<'a> {
let (mime_type, msg_type) = get_mime_type(mail)?;
let raw_mime = mail.ctype.mimetype.to_lowercase();
let filename = get_attachment_filename(mail);
info!(
self.context,
"add_single_part_if_known {:?} {:?}", mime_type, msg_type
"add_single_part_if_known {:?} {:?} {:?}", mime_type, msg_type, filename
);
let old_part_count = self.parts.len();
let is_attachment = mail
.get_content_disposition()?
.params
.iter()
.any(|(key, _value)| key.starts_with("filename"));
// regard `Content-Transfer-Encoding:`
match mime_type.type_() {
mime::TEXT | mime::HTML if !is_attachment => {
let decoded_data = match mail.get_body() {
Ok(decoded_data) => decoded_data,
Err(err) => {
warn!(self.context, "Invalid body parsed {:?}", err);
// Note that it's not always an error - might be no data
return Ok(false);
}
};
// check header directly as is_send_by_messenger is not yet set up
let is_msgrmsg = self.lookup_field("Chat-Version").is_some();
let mut simplifier = Simplify::new();
let simplified_txt = if decoded_data.is_empty() {
"".into()
} else {
let is_html = mime_type == mime::TEXT_HTML;
simplifier.simplify(&decoded_data, is_html, is_msgrmsg)
};
if !simplified_txt.is_empty() {
let mut part = Part::default();
part.typ = Viewtype::Text;
part.mimetype = Some(mime_type);
part.msg = simplified_txt;
part.msg_raw = Some(decoded_data);
self.do_add_single_part(part);
if let Ok(filename) = filename {
self.do_add_single_file_part(
msg_type,
mime_type,
&raw_mime,
&mail.get_body_raw()?,
&filename,
);
} else {
match mime_type.type_() {
mime::IMAGE | mime::AUDIO | mime::VIDEO | mime::APPLICATION => {
bail!("missing attachment");
}
mime::TEXT | mime::HTML => {
let decoded_data = match mail.get_body() {
Ok(decoded_data) => decoded_data,
Err(err) => {
warn!(self.context, "Invalid body parsed {:?}", err);
// Note that it's not always an error - might be no data
return Ok(false);
}
};
if simplifier.is_forwarded {
self.is_forwarded = true;
}
}
mime::TEXT
| mime::HTML
| mime::IMAGE
| mime::AUDIO
| mime::VIDEO
| mime::APPLICATION => {
// try to get file name from
// `Content-Disposition: ... filename*=...`
// or `Content-Disposition: ... filename*0*=... filename*1*=... filename*2*=...`
// or `Content-Disposition: ... filename=...`
// check header directly as is_send_by_messenger is not yet set up
let is_msgrmsg = self.lookup_field("Chat-Version").is_some();
let ct = mail.get_content_disposition()?;
let mut desired_filename = ct
.params
.iter()
.filter(|(key, _value)| key.starts_with("filename"))
.fold(String::new(), |mut acc, (_key, value)| {
acc += value;
acc
});
if desired_filename.is_empty() {
if let Some(param) = ct.params.get("name") {
// might be a wrongly encoded filename
desired_filename = param.to_string();
}
}
// if there is still no filename, guess one
if desired_filename.is_empty() {
if let Some(subtype) = mail.ctype.mimetype.split('/').nth(1) {
desired_filename = format!("file.{}", subtype,);
let mut simplifier = Simplify::new();
let simplified_txt = if decoded_data.is_empty() {
"".into()
} else {
return Ok(false);
let is_html = mime_type == mime::TEXT_HTML;
simplifier.simplify(&decoded_data, is_html, is_msgrmsg)
};
if !simplified_txt.is_empty() {
let mut part = Part::default();
part.typ = Viewtype::Text;
part.mimetype = Some(mime_type);
part.msg = simplified_txt;
part.msg_raw = Some(decoded_data);
self.do_add_single_part(part);
}
if simplifier.is_forwarded {
self.is_forwarded = true;
}
}
self.do_add_single_file_part(
msg_type,
mime_type,
&raw_mime,
&mail.get_body_raw()?,
&desired_filename,
);
_ => {}
}
_ => {}
}
// add object? (we do not add all objects, eg. signatures etc. are ignored)
@@ -651,6 +615,7 @@ impl<'a> MimeParser<'a> {
return;
}
};
info!(self.context, "added blobfile: {:?}", blob.as_name());
/* create and register Mime part referencing the new Blob object */
let mut part = Part::default();
@@ -902,12 +867,13 @@ pub struct Part {
pub param: Params,
}
/// return mimetype and viewtype for a parsed mail
fn get_mime_type(mail: &mailparse::ParsedMail<'_>) -> Result<(Mime, Viewtype)> {
let mimetype = mail.ctype.mimetype.parse::<Mime>()?;
let viewtype = match mimetype.type_() {
mime::TEXT => {
if !mailmime_is_attachment_disposition(mail) {
if !is_attachment_disposition(mail) {
match mimetype.subtype() {
mime::PLAIN | mime::HTML => Viewtype::Text,
_ => Viewtype::File,
@@ -941,14 +907,62 @@ fn get_mime_type(mail: &mailparse::ParsedMail<'_>) -> Result<(Mime, Viewtype)> {
Ok((mimetype, viewtype))
}
fn mailmime_is_attachment_disposition(mail: &mailparse::ParsedMail<'_>) -> bool {
if let Some(ct) = mail.ctype.params.get("Content-Disposition") {
return ct.to_lowercase().starts_with("attachment");
fn is_attachment_disposition(mail: &mailparse::ParsedMail<'_>) -> bool {
if let Ok(ct) = mail.get_content_disposition() {
return ct.disposition == DispositionType::Attachment
&& ct
.params
.iter()
.any(|(key, _value)| key.starts_with("filename"));
}
false
}
fn get_attachment_filename(mail: &mailparse::ParsedMail) -> Result<String> {
// try to get file name from
// `Content-Disposition: ... filename*=...`
// or `Content-Disposition: ... filename*0*=... filename*1*=... filename*2*=...`
// or `Content-Disposition: ... filename=...`
let ct = mail.get_content_disposition()?;
ensure!(
ct.disposition == DispositionType::Attachment,
"disposition not an attachment: {:?}",
ct.disposition
);
let mut desired_filename = ct
.params
.iter()
.filter(|(key, _value)| key.starts_with("filename"))
.fold(String::new(), |mut acc, (_key, value)| {
acc += value;
acc
});
println!("get_attachment_filename1: {:?}", desired_filename);
if desired_filename.is_empty() {
if let Some(param) = ct.params.get("name") {
// might be a wrongly encoded filename
desired_filename = param.to_string();
}
}
println!("get_attachment_filename2: {:?}", desired_filename);
// if there is still no filename, guess one
if desired_filename.is_empty() {
if let Some(subtype) = mail.ctype.mimetype.split('/').nth(1) {
desired_filename = format!("file.{}", subtype,);
} else {
bail!("could not determine filename: {:?}", ct.disposition);
}
}
println!("get_attachment_filename3: {:?}", desired_filename);
Ok(desired_filename)
}
// returned addresses are normalized.
fn get_recipients<S: AsRef<str>, T: Iterator<Item = (S, S)>>(headers: T) -> HashSet<String> {
let mut recipients: HashSet<String> = Default::default();
@@ -1053,6 +1067,29 @@ mod tests {
assert_eq!(recipients.len(), 2);
}
#[test]
fn test_is_attachment() {
let raw = include_bytes!("../test-data/message/mail_with_cc.txt");
let mail = mailparse::parse_mail(raw).unwrap();
assert!(!is_attachment_disposition(&mail));
let raw = include_bytes!("../test-data/message/mail_attach_txt.eml");
let mail = mailparse::parse_mail(raw).unwrap();
assert!(!is_attachment_disposition(&mail));
assert!(!is_attachment_disposition(&mail.subparts[0]));
assert!(is_attachment_disposition(&mail.subparts[1]));
}
#[test]
fn test_get_attachment_filename() {
let raw = include_bytes!("../test-data/message/html_attach.eml");
let mail = mailparse::parse_mail(raw).unwrap();
assert!(get_attachment_filename(&mail).is_err());
assert!(get_attachment_filename(&mail.subparts[0]).is_err());
let filename = get_attachment_filename(&mail.subparts[1]).unwrap();
assert_eq!(filename, "test.html")
}
#[test]
fn test_mailparse_content_type() {
let ctype =

View File

@@ -0,0 +1,25 @@
Chat-Disposition-Notification-To: tmp_6272287793210918@testrun.org
Subject: =?utf-8?q?Chat=3A_File_=E2=80=93_test=2Ehtml?=
Message-ID: Mr.XA6y3og8-az.WGbH9_dNcQx@testrun.org
Date: Sat, 07 Dec 2019 19:00:27 +0000
X-Mailer: Delta Chat Core 1.0.0-beta.12/DcFFI
Chat-Version: 1.0
To: <tmp_5890965001269692@testrun.org>
From: "=?utf-8?q??=" <tmp_6272287793210918@testrun.org>
Content-Type: multipart/mixed; boundary="mwkNRwaJw1M5n2xcr2ODfAqvTjcj9Z"
--mwkNRwaJw1M5n2xcr2ODfAqvTjcj9Z
Content-Type: text/plain; charset=utf-8
--
Sent with my Delta Chat Messenger: https://delta.chat
--mwkNRwaJw1M5n2xcr2ODfAqvTjcj9Z
Content-Type: text/html
Content-Disposition: attachment; filename="test.html"
Content-Transfer-Encoding: base64
PGh0bWw+PGJvZHk+dGV4dDwvYm9keT5kYXRh
--mwkNRwaJw1M5n2xcr2ODfAqvTjcj9Z--

View File

@@ -0,0 +1,40 @@
From holger@merlinux.eu Sat Dec 7 11:53:58 2019
Return-Path: <holger@merlinux.eu>
X-Original-To: holger+test@merlinux.eu
Delivered-To: holger+test@merlinux.eu
Received: from [127.0.0.1] (localhost [127.0.0.1])
(using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits))
(No client certificate requested)
by mail.merlinux.eu (Postfix) with ESMTPSA id 61825100531;
Sat, 7 Dec 2019 10:53:58 +0000 (UTC)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=merlinux.eu;
s=default; t=1575716038;
bh=xhaPssQzVOHRafcciTQqDnZ1Zi4GMwsmg9pHLH3i8P8=;
h=Date:From:To:Subject;
b=U4HxGDZ8RwLwRPFtIvRsb+x5BiyICnbbY2ZOGlZdLt12MuDTfiYi/phHiQUC402EY
GXb8dYgYr5+0PDiPBa7dyt2VQLC/h9QRfOA82tb1vpJYC+KksSAH0nYQqJvs7XrqCN
i95/jwZnsWrV7w72+xsrO5qPujIE68TmM5I9Cyec=
Received: by beto.merlinux.eu (Postfix, from userid 1000)
id 229D3820070; Sat, 7 Dec 2019 11:53:58 +0100 (CET)
Date: Sat, 7 Dec 2019 11:53:57 +0100
From: holger krekel <holger@merlinux.eu>
To: holger+test@merlinux.eu
Subject: hello
Message-ID: <20191207105357.GA6266@beto>
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="YiEDa0DAkWCtVeE4"
Content-Disposition: inline
--YiEDa0DAkWCtVeE4
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
siehe anhang
--YiEDa0DAkWCtVeE4
Content-Type: text/plain; charset=us-ascii
Content-Disposition: attachment; filename="x.txt"
hello
--YiEDa0DAkWCtVeE4--